A Comprehensive Review of Driver Drowsiness Detection Systems and Applied Technologies

Drowsiness among drivers is a significant contributing factor to traffic injuries, leading to a yearly toll of severe injuries and fatalities. This study reviews the literature in-depth on driving force sleepiness detection systems, considering physiological signals, facial expressions, and riding habits. It not only describes the current approaches for each class but also offers a comparative analysis of recently published studies, closely contrasting correctness, dependability, hardware specifications, and intrusiveness. The research also examines the advantages and disadvantages of each technique’s brilliance, highlighting the need for a hybrid machine that builds on each one’s advantages to guarantee performance, resilience, accuracy and real-time use. This crucial issue requires attention and creativity, as demonstrated by a different investigation looking for ways to reduce driving induce sleepiness by strengthening deeply ingrained patterns. This paper investigates several Convolution Neural network configurations, visual Geometry Group architectures (VGG16 and VGG19), a Generative antagonistic network (GAN), and a Multi-Layer Perceptron (MLP) and offers enhancing statistics to model real-world scenarios. Using eye photo datasets, general performance criteria including as accuracy, precision, F1-score, recall, and specificity are assessed. The findings highlight the ability for extensive makeup in fashions with accuracy and an F1-score of < 90%, as employing facts augmentation procedures led to an average improvement in accuracy and F1 rating of +4.3%.